Announcement

Collapse
No announcement yet.

New Stat - Baseball Prospectus: Deserved Runs Created

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    I was going to add something about how DRC+ may actually bring sabers and traditionalists closer together, because we're now all in the same boat, having to trust what some stathead with a computer says, rather than being able to verify it. But it seems that the implications of DRC+ go even further. Using that metric, it turns out--Say it Ain't So!--that Trout was not more valuable or productive than Cabrera in 2012 or 2013:

    https://www.baseballprospectus.com/n...ging-with-drc/

    According to this framework, Cabrera was not only a better hitter than Trout in 2012 and 2013, but better than him by enough to easily outstrip whatever edge Trout had in non-batting value. Cabrera had, according to our new math, a very narrow edge in total WARP over Trout in 2013, and a full win of cushion over him in 2012. Cabrera led MLB in DRC+ in both seasons, at 166 (21 points above 10th-place Trout, at 145) in 2012 and 187 (28 points better than third-place Trout, at 159) in 2013.
    They really tell the story of a player’s talent, though, when lumped into three categories:
    • Not-In-Play Runs: Runs gained or lost by the batter’s walks, strikeouts, and times hit by pitch, relative to what an average batter would have done in the same number of plate appearances;
    • In-Play Out Runs: Runs gained or lost by the outs the batter made on balls in play, relative to what an average batter would have done in the same number of plate appearances; and
    • Hit Runs: Runs gained or lost by the hits the batter got, relative to what an average batter would have done in the same number of plate appearances.

    These categories matter, because in them we see the three essential things a batter is trying to do at home plate: control the strike zone, hit the ball hard, and generate loft. Though by no means designed to do so, these could almost be considered surrogates for, say, contact rate, exit velocity, and launch angle, or for plate discipline, the hit tool, and the power tool, or for strikeout-to-walk ratio, BABIP, and isolated power. Whereas those are blunt instruments, though, these are scalpels. They tell us how much a hitter really does to get themselves on base, to avoid outs, and to drive the baseball, rather than just how often they manage to do those things.

    Here are Trout and Cabrera in 2012, through the lens of these three categories:
    Player NIP Runs Hit Runs BIP Out Runs Deserved Runs Above Average
    Mike Trout 0.5 23.8 14.2 38.0
    Miguel Cabrera 11.7 44.3 3.5 59.0
    Mongoose will love this:

    The DRC+ model gives us run values (above or below average) for the contributions of each hitter to all the possible outcomes of a plate appearance. It tells us, for instance, that Cabrera’s ability to consistently avoid strikeouts was worth 10.0 runs in 2012, while Trout’s above-average strikeout rate meant his whiffs cost the Angels 4.0 runs.
    Trout has made all the same improvements Cabrera made throughout his early and mid-20s. In 2015, he was the leader in DRC-based WARP for the first (and still only) time, and in both 2017 and 2018, he led the AL in DRC+ itself. This past season, for the first time, he truly was the best hitter in baseball, with a 180 DRC+. We were correct, as far back as 2012, to observe that Trout was doing something historic. We were wrong about what that was. Trout didn’t break baseball, and he wouldn’t be an inner-circle Hall of Famer if the productive phase of his career ended with 2018. Still, he’s traced an arc from elite young hitter to elite hitter, period, and if he has a normal aging curve from here, he’ll easily attain that all-time great status we’d already assigned to him under all the old offensive value frameworks.
    I gets even wilder when this analysis is performed on other seasons in the past decade:







    So, according to DRC+, while Trout was the deserving MVP not only in 2014, when he won it, and in 2015, when he finished second, Betts had a higher WARP in 2016, when Trout won it. It should be kept in mind, though, that BBPro's WARP, even before being modified by DRC+, frequently differed greatly from FG and BBRef WAR, because of differences in the way defense is evaluated. Martinez has a higher defense rating than Trout, which I don't understand at all.

    Some other complaint are emerging. Ty Cobb's 1922 season is now considered only worth 1.9 BWARP, and Craig Nettles was a better hitter than Rod Carew. Singles hitters seem to suffer in this system, because singles are considered far more a matter of luck than XBH.

    Everything we think we know about baseball right now is subject to debate and revision. As our numbers and methods advance by leaps and bounds, some of the sacrosanct ideas of sabermetrics—like Mike Trout’s perennial MVP candidacy, or the nonexistence of hot streaks—may end up being discarded. The sabermetric community ought to temper our numbers-driven certainty with the knowledge that we are always one breakthrough away from being wrong.
    https://www.baseballprospectus.com/n...tory-with-drc/
    Last edited by Stolensingle; 12-11-2018, 10:33 PM.

    Comment


    • #17
      Another point about DRC+ is that it's not only used to determined WARP, but also to modify traditional stats. Every player listed at BBPRo now has a "d" slash line as well as a traditional one. E.g., Betts's traditional slash line in 2018 was .346/.438/.640, while his d-slash was .331/.420/.597.

      https://legacy.baseballprospectus.co...this_year=2018

      Comment


      • #18
        I stumbled across this website this morning, haven't read the articles yet, but a deep dive of critiquing DRC is here...https://hareeb.com
        Jacquelyn Eva Marchand (1983-2017)
        http://www.tezakfuneralhome.com/noti...uelyn-Marchand

        Comment


        • #19
          Originally posted by Jar of Flies View Post
          I stumbled across this website this morning, haven't read the articles yet, but a deep dive of critiquing DRC is here...https://hareeb.com
          All the articles are worth a read, but this one might get to the heart of it best:
          https://hareeb.com/2018/12/10/cmon-m...s-drc-edition/

          While they continue to refine, we're waiting for something more meaningful in the meantime (just use other WAR measures for the time being and your best judgments).
          Jacquelyn Eva Marchand (1983-2017)
          http://www.tezakfuneralhome.com/noti...uelyn-Marchand

          Comment


          • #20
            That blog has had a series of analyses. He claims that Judge's evidence that DRC+ is more reliable, descriptive and predictive than wRC+ is flawed. He also claims it's biassed against pitchers. E.g., he did an analysis of all players with 1 PA and 1 walk. You would think their DRC+ values would be similar--there would be park and other situation effects, but still, you wouldn't expect major differences in DRC+ if the player had the same event in one PA. And yet the DRC+ values for position players on that list were very high, while those for pitchers were very low or even negative.

            Also, as I pointed out in the Trout thread in Current Events, the new DRC+ values result from major changes in park factors. The original DRC+ values were based on a very narrow range of park factors, so that Coors was not that much above average for hitters, and pitcher parks were almost neutral. Those factors have been changed so that they're more like the usual factors found at other sites. The DRC+ values are now not that much different from wRC+ values.

            I find it very interesting that Judge has not responded to Hareeb--neither in the ocmments to his blog, nor in an update that Judge published a few days ago.
            Last edited by Stolensingle; 02-10-2019, 05:42 PM.

            Comment


            • #21
              Originally posted by Stolensingle View Post
              That blog has had a series of analyses. He claims that Judge's evidence that DRC+ is more reliable, descriptive and predictive than wRC+ is flawed. He also claims it's biassed against pitchers. E.g., he did an analysis of all players with 1 PA and 1 walk. You would think their DRC+ values would be similar--there would be park and other situation effects, but still, you wouldn't expect major differences in DRC+ if the player had the same event in one PA. And yet the DRC+ values for position players on that list were very high, while those for pitchers were very low or even negative.

              Also, as I pointed out in the Trout thread in Current Events, the new DRC+ values result from major changes in park factors. The original DRC+ values were based on a very narrow range of park factors, so that Coors was not that much above average for hitters, and pitcher parks were almost neutral. Those factors have been changed so that they're more like the usual factors found at other sites. The DRC+ values are now not that much different from wRC+ values.

              I find it very interesting that Judge has not responded to Hareeb--neither in the ocmments to his blog, nor in an update that Judge published a few days ago.
              So they corrected the one thing that made it stand-out from the others the most? Not much to see anymore then.

              This article discusses the same regression flaws that DRA has for pitchers and how BBPRo can say that Niekro, Palmer, and Glavine were all actually below average pitchers in terms of run prevention.

              I have tinkered with taking career regression and applying it each year as an average. But that is quite different than taking yearly regression and applying it to a career. It completely ignores bayesian rules of increased sample size and priors, etc.

              But, man, is BBPro light years ahead of everyone else in terms of catcher defense. Defintilat their calling-card right now.
              Last edited by Bothrops Atrox; 02-10-2019, 06:09 PM.
              1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

              1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

              1996 2000 2001 2002 2005 2009 2012 2014 2015


              The Top 100 Pitchers In MLB History
              The Top 100 Position Players In MLB History

              Comment


              • #22
                Of course, there will be a newer stat created by the agents--Deserved $$$ created.

                Comment


                • #23
                  Originally posted by Bothrops Atrox View Post

                  So they corrected the one thing that made it stand-out from the others the most? Not much to see anymore then.
                  When I saw what Judge did, I couldn't help wondering if he felt that reducing the difference would allay all the criticism he was getting. People would see that Arenado was as valuable as Trout, and say, WTF? Now that his value has been reduced, people will stop complaining.

                  I have tinkered with taking career regression and applying it each year as an average. But that is quite different than taking yearly regression and applying it to a career. It completely ignores bayesian rules of increased sample size and priors, etc.
                  That is exactly what I have criticized Lionel for in his new WAR system.



                  Comment


                  • #24
                    Originally posted by Stolensingle View Post

                    When I saw what Judge did, I couldn't help wondering if he felt that reducing the difference would allay all the criticism he was getting. People would see that Arenado was as valuable as Trout, and say, WTF? Now that his value has been reduced, people will stop complaining.



                    That is exactly what I have criticized Lionel for in his new WAR system.


                    Judge is a man of pride and integrity. If he changed, I would bet it is because he felt correct data called for it and not just to appease the majority. He has always stood bye his past results regardless of how they were percieved. At least that is my initial guess.
                    1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

                    1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

                    1996 2000 2001 2002 2005 2009 2012 2014 2015


                    The Top 100 Pitchers In MLB History
                    The Top 100 Position Players In MLB History

                    Comment


                    • #25
                      Originally posted by Bothrops Atrox View Post

                      Judge is a man of pride and integrity. If he changed, I would bet it is because he felt correct data called for it and not just to appease the majority. He has always stood bye his past results regardless of how they were percieved. At least that is my initial guess.
                      I agree that he wouldn't change the data to mollify critics, certainly he wouldn't do that. But if he hadn't come under such heavy criticism for park factors, I don't think he would have gone back to take another look at them.

                      He originally said that the differences were because he used one year data. But the differences between his park factors and the ones of other sites seemed to exist for every year of the past several years, so there has to be something else going on. E.g., if his factor for Coors was 1.04 in 2014, 2015, 2016, 2017 and 2018, then the 2018 figure at other sites, based on three year or five year data, should also be 1.04, when in fact it's more like 1.15-1.20.

                      In his update, he says:

                      There was never anything wrong with the park ratings themselves, just the way those effects were being isolated from batter predictions. The batter predictions are now 100 percent park-isolated.
                      I think what he means is that the park factors are treated as completely independent of all the other factors he looks at. And somehow, by doing that, there is a dramatic change in the park factors.

                      Your favorite Colorado hitters will now be adjusted a bit downward from where they were before, and your hitters from parks that suppress offense are now being treated more fairly.
                      A bit? Arenado goes from 146 to 137. More fairly? Trout goes from 145 to 160 in 2012, 159 to 169 in 2013, and 155 to 167 in his career. I think some of that change must involve other factors than park, but it makes you wonder how you can get so much fluctuation by tinkering with other factors.
                      Last edited by Stolensingle; 02-10-2019, 07:45 PM.

                      Comment


                      • #26
                        Originally posted by Stolensingle View Post

                        I agree that he wouldn't change the data to mollify critics, certainly he wouldn't do that. But if he hadn't come under such heavy criticism for park factors, I don't think he would have gone back to take another look at them.
                        Yeah, I agree with that.
                        1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

                        1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

                        1996 2000 2001 2002 2005 2009 2012 2014 2015


                        The Top 100 Pitchers In MLB History
                        The Top 100 Position Players In MLB History

                        Comment

                        Ad Widget

                        Collapse
                        Working...
                        X